PARAMETRIC IMAGE ESTIMATION IN ANALYZER-BASED PHASE CONTRAST IMAGING
Since the discovery of X-rays, by W.E. Roentgen in 1895, absorption-contrast imaging has been widely used in medical diagnostic, for example: mammography, chest X-ray, angiography etc. Over the last 25 years many researchers have shown that phase-contrast X-ray imaging is capable of obtaining a better contrast then classical absorption-contrast imaging. This is especially true in soft tissue imaging like mammography or at higher Xrays energies. During the past five years a significant effort has made in development of a tabletop analyzer-based phase-contrast system. In this system a perfect crystal is used to generate a quasi-monochromatic beam, which after interacting with an object is analyzed by a second perfect crystal called analyzer. The need of quasi-monochromatic beam and the low brilliance of conventional X-ray sources (CXS) has been the major limiting factor for tabletop systems. The work presents in this thesis aims to develop novel reconstruction methods for tabletop analyzer-based phase-contrast imaging (ABI) systems. The presented reconstruction methods goal is to minimize the impact of CXS low brilliance by utilizing phase-contrast image formation model and maximum or maximum-a-posteriori Poisson likelihood approach. Finally, a fast-convergent conjugate gradient optimization algorithm has been derived specifically for Poisson likelihood function maximization.